Posted by Sten Westgard, MS
[Hat-tip to the AACC Point-of-Care listserve, which first posted a notice about this article]
The Pennsylvania Patient Safety Advisory has a regular electronic newsletter highlighting new science and studies about healthcare safety. Their December 2011 issue has a particularly interesting article for laboratory testing:
Point-of-Care Technology: Glucose Meter's Role in Patient Care, Lea Anne Gardner, PhD, RN, Senior Patient Safety Analyst, Pennsylvania Patient Safety Authority.
This review examined more than 1,300 reports of glucose-meter problems from the Pennsyvlania reporting system database from 2004 to 2011. Of those reports, 71 near-miss or adverse event reports occurred. Most intriguing are the report excerpts directly quoted in the study. Of those reports, 72% of the near-miss or adverse events occurred with high-blood glucose results. That is, where the glucose meter had a sudden high value that may or may not have been reflective of the actual patient's clinical state. For example:
"A patient's blood sugar was checked using a [glucose meter]. The lunchtime result was 517. A [blood glucose test] was [immediately] retaken to check for accuracy, and the result was greater than 600. A blood [laboratory] test was conducted per protocol, and the [lab] glucose [result] was 136..."
What do you think happened next?
"The nurse used the serum glucose as the actual result and reported a malfunction in the machine to the lab and the [nurse manager]. The [nurse manager] was made aware of the situation and took the glucose meter out of use on the floor and notified the supervisor in the lab. The supervisor in lab removed the machine from service and replaced the machine with a new machine."
In this particular case, it's troubling that the patient becomes the quality control check. The health system didn't catch the error before it affected patient results - instead they essentially used the patient results to determine if something was wrong with the device. With waived tests, the users only need to follow the manufacturer's instructions, not adhere to the usual (more rigorous) QC practices found in the laboratory. So some of the checks that might have caught this earlier may not have been in place.
"A [glucose meter] gave a reading of 468, and the physician was notified and ordered insulin, which was given. The morning [fasting blood sugar] wfrom lab came back with result of 122. A [glucose meter test] was redone on a different machine with a reading of 135. [The second glucose meter] was calibrated [within normal limits] following the discrepancy. [The initial glucose meter] machine was taken out of service. The patient's [blood sugar] was check [on the hour for six hours] and observed for hypoglycemia. No adverse outcome was observed."
The Patient Safety Authority states, with our emphasis added here: "Before treating a patient's high glucose level, especially a new high glucose level, the result needs to be validated."
"Patient's glucose meter reading [prior to lunch] read "HI." Patient covered with 6 units of Novolog insulin. A stat venous draw was done; [the result] was 62. The patient was lethargic at this time and was given orange juice with sugar. [Repeat glucose meter result] came up to 156."
The review points out a specific FDA warning letter from 2009:
"[M]any laboratories continue to use SMBG [test strip based] devices, cleared only as aids in the management of diabetic patients, in these settings, even though they are not FDA cleared to diagnose disease or to maintain tight glycemic control of diabetic and non-diabetic patients in teh hospital environment. This practice can be problematic."
In other words, using test strip based glucose meters as definitive tests is an off-label use. They are aids in monitoring, but as soon as the monitor picks up something interesting or important, you need to validate it with a more accurate method.
The Patient Safety Authority provides a list of 12 steps that hospital policies should adopt to ensure appropriate use of blood glucose meters, including daily testing to explicitly compare the comparisons of glucose meters and lab serum results.
Here is the conclusion from the Patient Safety Authority:
"Blood glucose management in hospitalized patients is a multifacted process that requires critical thinking. Test results are one important measure of a patient's condition but should never be used as the sole basis for treatment. Treating a patient based on measurements along can lead to serious and fatal patient events. Blood glucose meter results provide a starting point of inquiry about a patient's health status."
This actually isn't news to the laboratory, I think. There are plenty of studies out there discussing the performance short-comings of glucose POC devices. Even the marketing and instruction for use on most glucose meters will state something to this effect. The real problem is that clinician's have gotten so comfortable with glucose meter results that they are strating to go off-label in their use. Labs and those who deal with glucose meters need to make sure that the meter limitations are always kept in mind.
Otherwise, something like this can happen.
Comments